This application provides an image recognition method and device. The image recognition method includes: setting a presentation time sequence corresponding to an image sequence includes N images, the presentation time sequence includes unequal presentation times, a difference between any two presentation times of the unequal presentation times is k×Δ, k is a positive integer, and Δ is a preset time period value; processing the image sequence by using a computer vision algorithm, to obtain a computer vision signal corresponding to each image in the image sequence; obtaining a feedback signal that is corresponding to each image in the image sequence generated when an observation object watches the image sequence displayed in the presentation time sequence; and fusing, for each image in the image sequence, a corresponding computer vision signal and a corresponding feedback signal to obtain a target recognition signal of each image in the image sequence.
Legal claims defining the scope of protection. Each claim is shown in both the original legal language and a plain English translation.
1. A brain-computer combination image recognition method based on image sequence presentation, comprising: setting a presentation time sequence corresponding to an image sequence, wherein the image sequence comprises N images, N is a positive integer, the presentation time sequence comprises a presentation time of an image i, the presentation time of the image i is determined based on a duration impact parameter that comprises at least one of a first recognition probability calculated using a computer vision algorithm or a fatigue state parameter corresponding to the image i, and is to indicate a time period from a presentation start moment of the image i to a presentation start moment of a next adjacent image, the image i is any image in the image sequence, the presentation time sequence comprises at least two unequal presentation times, a difference between any two presentation times of the at least two unequal presentation times is k×Δ, k is a positive integer, and Δ is a preset time period value; processing the image sequence by using the computer vision algorithm, to obtain a computer vision signal corresponding to the image i in the image sequence; obtaining a feedback signal that is generated when an observation object watches the image sequence displayed in the presentation time sequence and that corresponds to the image i in the image sequence, wherein the feedback signal is to indicate a reaction of the observation object to the watched image; and fusing, for the image i in the image sequence, the computer vision signal corresponding to the image i and the feedback signal corresponding to the image i to obtain a target recognition signal of the image i in the image sequence for image recognition.
This invention relates to brain-computer interface (BCI) and computer vision technologies for image recognition. It addresses the challenge of optimizing image presentation for improved recognition accuracy by considering both computational analysis and human observation. The method involves presenting a sequence of N images. A presentation time sequence is established for these images. The duration for which each image (image i) is displayed is determined by a duration impact parameter. This parameter incorporates at least one of a first recognition probability calculated by a computer vision algorithm or a fatigue state parameter associated with image i. The presentation time for image i defines the interval from its display start to the display start of the next adjacent image. The presentation time sequence includes at least two different display durations, where the difference between any two durations is a multiple (k) of a preset time period (Δ). The process includes using a computer vision algorithm to analyze the image sequence and generate a computer vision signal for each image. Simultaneously, a feedback signal is captured from an observation object watching the images as they are presented according to the determined time sequence. This feedback signal reflects the observer's reaction to each image. Finally, for each image, the computer vision signal and the feedback signal are fused to produce a target recognition signal, which is then used for image recognition.
2. The method according to claim 1 , wherein before the setting of the presentation time sequence corresponding to the image sequence, the method further comprises: receiving M images from a camera device, wherein M is an integer greater than 1; and selecting N images from the M images as the image sequence, wherein N is less than or equal to M.
This invention relates to image processing, specifically methods for selecting and sequencing images from a camera device to optimize presentation timing. The problem addressed is the need to efficiently capture and select a subset of images from a continuous stream to form a coherent sequence, ensuring optimal presentation timing for applications such as video playback, surveillance, or real-time monitoring. The method involves receiving M images from a camera device, where M is an integer greater than 1. From these M images, N images are selected to form an image sequence, with N being less than or equal to M. This selection process ensures that only the most relevant or high-quality images are chosen for further processing. Before setting the presentation time sequence for the selected image sequence, the method includes this preliminary step of image selection. The presentation time sequence determines how the selected images are displayed or processed in a timed manner, ensuring smooth and synchronized playback or analysis. The invention improves efficiency by reducing the number of images processed while maintaining temporal coherence, which is critical for applications requiring real-time or near-real-time image analysis. The selection criteria for choosing N images from M may be based on factors such as image quality, motion detection, or user-defined parameters. This method ensures that the final image sequence is optimized for the intended application, whether for display, storage, or further analysis.
3. The method according to claim 1 , wherein the first recognition probability is to indicate a probability that the image i comprises a preset image feature; wherein the fatigue state parameter is to indicate a fatigue degree of the observation object when the observation object observes the image i; and wherein the presentation time is inversely correlated with the first recognition probability, and positively correlated with the fatigue state parameter.
This invention relates to an adaptive image presentation system designed to optimize visual engagement by dynamically adjusting display times based on image recognition and observer fatigue. The system addresses the problem of maintaining user attention during prolonged image viewing sessions by preventing fatigue-induced performance degradation. The method involves analyzing an image to determine a recognition probability, which quantifies the likelihood that the image contains a predefined feature. Simultaneously, the system assesses the observer's fatigue state, which reflects their cognitive load or visual fatigue during observation. The presentation time for the image is then calculated using an inverse relationship with the recognition probability—images with higher recognition likelihood are displayed for shorter durations—and a direct relationship with the fatigue state parameter—images are shown longer when the observer exhibits higher fatigue. This adaptive approach ensures that images are presented in a manner that balances recognition efficiency and observer comfort, enhancing overall engagement and reducing fatigue-related errors. The system may integrate with various display technologies, including augmented reality or digital signage, to provide real-time adjustments based on continuous monitoring of both image content and observer state.
4. The method according to claim 3 , wherein the duration impact parameter comprises the fatigue state parameter, and before the determining of the presentation time for the image i in the image sequence based on the duration impact parameter, the method further comprises: predicting, according to a fatigue rule, the fatigue state parameter corresponding to the image i in the image sequence, wherein the fatigue rule is to indicate a change rule of a fatigue degree of the observation object.
This invention relates to a method for optimizing image presentation in a sequence to account for observer fatigue. The problem addressed is the degradation of visual perception over time due to fatigue, which can reduce the effectiveness of image-based systems such as medical imaging, surveillance, or augmented reality. The method predicts the fatigue state of an observer to dynamically adjust the presentation time of each image in a sequence, ensuring optimal perception and reducing errors caused by fatigue. The method first predicts a fatigue state parameter for each image in the sequence using a predefined fatigue rule. This rule models how an observer's fatigue level changes over time, considering factors like exposure duration and image content. The fatigue state parameter quantifies the observer's fatigue level at the time of viewing each image. The presentation time for each image is then determined based on this parameter, allowing for longer or shorter display times depending on the predicted fatigue state. This adaptive approach ensures that images are presented when the observer is least fatigued, improving accuracy and reducing cognitive load. The method may also incorporate additional duration impact parameters, such as the complexity or contrast of the image, to further refine the presentation timing. By dynamically adjusting display times, the system enhances the reliability of image-based decision-making in applications where observer fatigue is a critical factor.
5. The method according to claim 3 , wherein the obtaining of a feedback signal that is generated when the observation object watches the image sequence displayed in the presentation time sequence and that corresponds to the image i in the image sequence comprises: in a process of displaying the image sequence in the presentation time sequence, obtaining a fatigue state parameter corresponding to an image j; adjusting, based on the fatigue state parameter corresponding to the image j, a presentation time of the presentation time sequence, wherein the image i corresponds to an image to be displayed after the image j in the image sequence, wherein the image j is any image in the image sequence.
This invention relates to a system for monitoring and adjusting the presentation of an image sequence based on viewer fatigue. The problem addressed is the need to dynamically adapt the display of visual content to prevent viewer fatigue, which can degrade attention and comprehension. The system obtains a feedback signal from a viewer as they watch an image sequence presented in a specific time sequence. This feedback signal corresponds to individual images in the sequence and is used to assess the viewer's fatigue state. When a fatigue state parameter is detected for a given image (referred to as image j), the system adjusts the presentation time of subsequent images (referred to as image i) in the sequence. The adjustment ensures that the display timing is optimized to reduce fatigue, improving the viewing experience. The fatigue state parameter can be derived from physiological signals, eye-tracking data, or other biometric indicators. The system dynamically modifies the presentation time sequence to maintain optimal engagement, particularly useful in applications like medical imaging, training simulations, or extended viewing sessions. The invention ensures that the display adapts in real-time to the viewer's condition, enhancing both comfort and effectiveness.
6. The method according to claim 5 , wherein the obtaining of the fatigue state parameter corresponding to the image j comprises: obtaining the fatigue state parameter based on fatigue state information that is sent by a sensor and that is obtained when the observation object watches the image j.
This invention relates to monitoring fatigue states of individuals observing visual content, such as images or videos. The problem addressed is the need for accurate and real-time assessment of fatigue levels in observers to improve safety, productivity, or user experience in applications like driver monitoring, workplace safety, or entertainment systems. The method involves obtaining a fatigue state parameter for an observer watching a specific image or video frame. This parameter is derived from fatigue state information collected by sensors while the observer views the image. The sensors may include eye-tracking devices, facial recognition cameras, or physiological monitors that detect signs of fatigue, such as eye closure rates, pupil dilation, or facial muscle relaxation. The collected data is processed to quantify the observer's fatigue level, which can then be used to trigger alerts, adjust content delivery, or modify system behavior to mitigate risks associated with fatigue. The method ensures that fatigue assessment is dynamic and context-aware, adapting to the observer's real-time state. This approach enhances applications where fatigue detection is critical, such as autonomous vehicle safety systems, medical monitoring, or fatigue management in high-stakes environments. The invention improves upon prior art by integrating sensor-based fatigue detection directly into the content observation process, providing more precise and timely feedback.
7. The method according to claim 3 , wherein the determining of the presentation time for the image i in the image sequence based on the duration impact parameter comprises: for the image i in the image sequence, finding a presentation time corresponding to the duration impact parameter from a first mapping table, wherein the first mapping table comprises a plurality of duration impact parameters and presentation times respectively corresponding to the plurality of duration impact parameters.
This invention relates to image sequence processing, specifically determining optimal presentation times for individual images within a sequence based on their impact on perceived duration. The problem addressed is ensuring consistent or controlled visual perception of time when displaying image sequences, such as in video playback or animation, where certain images may distort the viewer's sense of elapsed time. The method involves analyzing each image in a sequence to calculate a duration impact parameter, which quantifies how that image affects the perceived duration of the sequence. For each image, the method then determines its presentation time by consulting a predefined mapping table. This table contains multiple duration impact parameters paired with corresponding presentation times, allowing the system to select an appropriate display duration for each image based on its calculated impact. The mapping table ensures that images with higher duration impact parameters are displayed for longer or shorter durations as needed to maintain temporal consistency or achieve a desired perceptual effect. This approach enables dynamic adjustment of image display times to compensate for variations in visual content, improving the accuracy of time perception in image sequences. The method is particularly useful in applications where precise temporal control is required, such as in medical imaging, scientific visualization, or entertainment media.
8. The method according to claim 3 , further comprising: in response detecting that the fatigue state parameter obtained when the observation object observes an image q is greater than or equal to a first fatigue threshold, controlling to stop displaying images to be displayed after the image q in the image sequence, and obtaining an image with a recognition probability that is greater than or equal to a first probability threshold in the images to be displayed after the image q, wherein the image q is any image in the image sequence; and in response to detecting that the fatigue state parameter of the observation object is less than or equal to a second fatigue threshold, controlling to sequentially display the image.
This invention relates to a system for monitoring and managing observer fatigue during image observation tasks, particularly in applications such as medical imaging, security surveillance, or quality inspection. The problem addressed is the degradation of observation accuracy due to observer fatigue, which can lead to missed critical details or errors in judgment. The method involves tracking a fatigue state parameter of an observer while they view a sequence of images. When the fatigue state parameter exceeds a first threshold, the system pauses the display of subsequent images and selects an image from the remaining sequence that has a recognition probability above a predefined threshold. This ensures that high-priority or critical images are not overlooked due to fatigue. Once the observer's fatigue state falls below a second threshold, the system resumes the sequential display of images. The fatigue state parameter may be derived from physiological signals, eye-tracking data, or behavioral metrics. The recognition probability of an image is determined based on factors such as image complexity, observer expertise, or historical performance data. This adaptive approach optimizes observation efficiency while maintaining accuracy, particularly in high-stakes environments where sustained attention is critical.
9. The method according to claim 3 , wherein the fusing, for the image i in the image sequence, of the computer vision signal and the feedback signal to obtain the target recognition signal of the image i in the image sequence comprises: determining, for the image i in the image sequence based on at least one of the first recognition probability, the fatigue state parameter, or the presentation time, a first weight corresponding to the image i in the image sequence, wherein the first weight is a weight used when the feedback signal is to determine the target recognition signal, the first weight is inversely correlated with the first recognition probability, the first weight is inversely correlated with the fatigue state parameter, and the first weight is positively correlated with the presentation time; and fusing, for the image i in the image sequence based on a corresponding first weight, the computer vision signal and a the feedback signal to obtain the target recognition signal of the image i in the image sequence.
This invention relates to a method for improving target recognition in computer vision systems by dynamically fusing computer vision signals with feedback signals, particularly in scenarios where human or system fatigue may affect performance. The method addresses the problem of maintaining accurate target recognition over time, especially when recognition confidence (first recognition probability) decreases or when fatigue (fatigue state parameter) increases, by adjusting the influence of feedback signals based on these factors. For each image in a sequence, the method determines a first weight that inversely correlates with both the recognition probability and the fatigue state parameter, meaning lower confidence or higher fatigue reduces the weight. Conversely, the weight positively correlates with presentation time, meaning longer exposure to the target increases the weight. This weight is then used to fuse the computer vision signal (e.g., raw sensor data) with the feedback signal (e.g., human or system corrections) to produce a more reliable target recognition signal. The fusion process dynamically balances the contributions of the original computer vision output and the feedback, ensuring robustness against fatigue and recognition errors. The method is particularly useful in applications like autonomous systems, surveillance, or medical imaging where sustained accuracy is critical.
10. The method according to claim 1 , wherein the fusing, for the image i in the image sequence, of the computer vision signal and the feedback signal to obtain the target recognition signal of the image i in the image sequence comprises: fusing, for the image i in the image sequence, the computer vision signal, the feedback signal, and at least one additional feedback signal corresponding to the image to obtain the target recognition signal of the image i in the image sequence.
This invention relates to computer vision systems that improve target recognition by fusing multiple feedback signals with computer vision outputs. The problem addressed is the limited accuracy of standalone computer vision models, which often fail to adapt to dynamic environments or account for real-world variability. The solution involves integrating feedback signals from multiple sources to enhance recognition performance. The method processes an image sequence, where each image undergoes computer vision analysis to generate a computer vision signal. This signal is then combined with a primary feedback signal, which may come from user input, sensor data, or other external sources. Additionally, at least one extra feedback signal is incorporated to refine the recognition process. The fusion of these signals—computer vision output, primary feedback, and supplementary feedback—produces a target recognition signal for each image. This multi-source fusion improves accuracy by leveraging diverse data inputs, reducing errors, and adapting to changing conditions. The approach is particularly useful in applications like autonomous navigation, surveillance, or quality control, where real-time adaptation is critical. The invention ensures robust target recognition by dynamically adjusting to feedback from multiple sources, enhancing reliability in complex scenarios.
11. The method according to claim 10 , wherein the fatigue state parameter comprises at least two fatigue state parameters respectively generated when at least two observation objects observe a same image.
This invention relates to systems for assessing fatigue states of observers, particularly in scenarios where multiple individuals observe the same visual content. The technology addresses the problem of accurately determining fatigue levels in observers to improve safety, performance, or user experience in applications such as driving, surveillance, or medical monitoring. The method involves generating at least two fatigue state parameters for a single image by analyzing the physiological or behavioral responses of at least two different observation objects (e.g., individuals or sensors) as they observe the same image. These fatigue state parameters may include metrics such as eye movement patterns, blink rates, pupil dilation, or other indicators of cognitive or physical fatigue. By comparing or aggregating these parameters, the system can assess how fatigue affects perception or decision-making across different observers. This approach helps identify inconsistencies or trends in fatigue responses, enabling adaptive adjustments in real-time, such as alerting users, modifying display settings, or triggering interventions. The method may also involve preprocessing the image or observer data to enhance accuracy, such as normalizing lighting conditions or filtering noise. The system can be applied in environments where multiple observers interact with shared visual content, ensuring reliable fatigue monitoring for safety-critical or high-performance applications.
12. The method according to claim 1 , wherein the computer vision signal is the first recognition probability determined by using the computer vision algorithm; before the fusing, for the image i in the image sequence, of the computer vision signal and the feedback signal to obtain the target recognition signal of the image i in the image sequence, the method further comprises: calculating, for the image i in the image sequence, a second recognition probability of the image i in the image sequence based on the feedback signal, wherein the second recognition probability is to indicate a probability that the observation object determines that the image i comprises the preset image feature; and wherein the fusing, for the image i in the image sequence, the computer vision signal and the feedback signal to obtain the target recognition signal of the image i in the image sequence comprises: calculating, for the image i in the image sequence, a target recognition probability of the image i in the image sequence based on the first recognition probability and the second recognition probability.
The invention relates to computer vision systems that analyze image sequences to detect specific features. The core problem addressed is improving recognition accuracy by combining automated computer vision analysis with human feedback. Traditional computer vision algorithms may produce unreliable results due to variations in lighting, occlusion, or other environmental factors. The invention enhances recognition by integrating a feedback signal from human observers with the computer vision output. The method involves processing an image sequence where each image is analyzed by a computer vision algorithm to generate a first recognition probability indicating the likelihood that a preset image feature is present. Additionally, a second recognition probability is calculated based on feedback from human observers, reflecting their assessment of whether the image contains the preset feature. These probabilities are then fused to produce a target recognition probability for each image, improving overall accuracy. The fusion process combines the automated and human-derived probabilities to generate a more reliable target recognition signal. This approach leverages both computational efficiency and human judgment to enhance feature detection in image sequences.
13. An image recognition device, comprising: a processor; a memory; and an interface circuit, wherein the memory, the interface circuit and the processor are interconnected, wherein the memory stores program instructions, which, when executed by the processor, cause the processor to: set a presentation time sequence corresponding to an image sequence, wherein the image sequence comprises N images, N is a positive integer, the presentation time sequence comprises a presentation time of an image i, the presentation time of the image i is determined based on a duration impact parameter that comprises at least one of a first recognition probability calculated using a computer vision algorithm or a fatigue state parameter corresponding to the image i, and is to indicate a time period from a presentation start moment of the image i to a presentation start moment of a next adjacent image, the image is any image in the image sequence, the presentation time sequence comprises at least two unequal presentation times, a difference between any two presentation times of the at least two unequal presentation times is k×Δ, k is a positive integer, and Δ is a preset time period value, process the image sequence by using the computer vision algorithm, to obtain a computer vision signal corresponding to the image i in the image sequence, cause the interface circuit to obtain a feedback signal that is generated when an observation object watches the image sequence displayed in the presentation time sequence and that corresponds to the image i in the image sequence, wherein the feedback signal is to indicate a reaction of the observation object to the watched image, and fuse, for the image i in the image sequence, the computer vision signal and the feedback signal to obtain the target recognition signal of the image i in the image sequence for image recognition.
This invention relates to an image recognition device designed to optimize the presentation and analysis of image sequences for improved recognition accuracy and user engagement. The device addresses the challenge of balancing recognition performance with user fatigue by dynamically adjusting the display duration of each image in a sequence based on both computational and physiological factors. The device includes a processor, memory, and interface circuit interconnected to process an image sequence comprising N images. A presentation time sequence is set for the images, where each image's display duration is determined by a duration impact parameter. This parameter includes either a first recognition probability (calculated via a computer vision algorithm) or a fatigue state parameter (indicating the observer's fatigue level). The presentation times vary, with differences between any two unequal times being a multiple (k) of a preset time period (Δ). The processor processes the image sequence using a computer vision algorithm to generate a computer vision signal for each image. The interface circuit captures a feedback signal from the observer, reflecting their reaction to the displayed image. These signals are fused to produce a target recognition signal for each image, enhancing recognition accuracy by combining computational analysis with real-time observer feedback. The system ensures adaptability to both image complexity and observer state, improving efficiency and user experience in image recognition tasks.
14. The image recognition device according to claim 13 , wherein: the interface circuit is configured to receive M images from a camera device, wherein M is an integer greater than 1; and the processor further is configured to select N images from the M images as the image sequence, wherein N is less than or equal to M.
The invention relates to an image recognition device designed to process multiple images from a camera to improve recognition accuracy. The device addresses the challenge of efficiently selecting relevant images from a larger set to enhance performance in tasks such as object detection or scene analysis. The device includes an interface circuit that receives M images from a camera, where M is an integer greater than 1. A processor within the device then selects N images from the M images to form an image sequence, with N being less than or equal to M. This selection process helps filter out redundant or low-quality images, ensuring only the most relevant frames are used for further analysis. The device may also include a memory for storing the selected images and a recognition circuit for performing image recognition tasks on the filtered sequence. The selection criteria for choosing N images from M may involve factors such as image quality, motion detection, or other relevance metrics to optimize recognition performance. This approach reduces computational overhead and improves the efficiency of image recognition systems in applications like surveillance, autonomous vehicles, or industrial inspection.
15. The image recognition device according to claim 13 , wherein the first recognition probability is to indicate a probability that the image i comprises a preset image feature, the fatigue state parameter is to indicate a fatigue degree of the observation object when the observation object observes the image i, the presentation time is inversely correlated with the first recognition probability, and the presentation time is positively correlated with the fatigue state parameter.
The invention relates to an image recognition device designed to optimize the presentation of images based on recognition probability and observer fatigue. The device addresses the problem of efficiently displaying images to an observer while accounting for their cognitive load and fatigue levels. The device includes an image recognition unit that determines a first recognition probability, representing the likelihood that an image contains a preset feature. A fatigue state detection unit assesses the observer's fatigue level, generating a fatigue state parameter. The device adjusts the presentation time of the image based on these factors: the presentation time decreases as the recognition probability increases (since easier-to-recognize images require less time) and increases as the fatigue state parameter rises (to accommodate higher fatigue levels). This dynamic adjustment ensures optimal image display efficiency while minimizing observer strain. The device may also include a second recognition unit that further analyzes the image for additional features, refining the presentation strategy. The overall system enhances user experience by balancing recognition accuracy with fatigue management, particularly useful in applications requiring prolonged visual attention, such as surveillance or medical imaging.
16. The image recognition device according to claim 15 , wherein the duration impact parameter comprises the fatigue state parameter, and the image recognition device further comprises a prediction unit; and wherein the processor is configured to predict, according to a fatigue rule, the fatigue state parameter corresponding to the image i in the image sequence, wherein the fatigue rule is to indicate a change rule of a fatigue degree of the observation object.
This invention relates to image recognition devices that analyze sequences of images to assess the fatigue state of an observation object, such as a person or machine component. The device addresses the challenge of accurately detecting fatigue over time by incorporating a fatigue state parameter as part of a duration impact parameter, which influences image recognition outcomes. The system includes a prediction unit that estimates the fatigue state of the observation object in each image based on a predefined fatigue rule. This rule defines how fatigue levels change over time, allowing the device to dynamically adjust recognition processes to account for fatigue-related variations in appearance or behavior. The processor uses this predicted fatigue state to improve recognition accuracy, ensuring reliable performance even as fatigue progresses. The invention enhances existing image recognition systems by integrating temporal fatigue modeling, making it particularly useful in applications like driver monitoring, industrial equipment inspection, or medical diagnostics where fatigue detection is critical. The system dynamically adapts to fatigue progression, reducing errors caused by fatigue-induced changes in the observed object.
17. The image recognition device according to claim 15 , wherein the interface circuit is further configured to: in a process of displaying the image sequence in the presentation time sequence, obtain a fatigue state parameter corresponding to an image j; and adjust, based on the fatigue state parameter corresponding to the image j, a presentation time that is in the presentation time sequence, wherein the image i corresponds to an image to be displayed after the image j in the image sequence, wherein the image j is any image in the image sequence.
This invention relates to an image recognition device designed to enhance user engagement by dynamically adjusting the display duration of images in a sequence based on viewer fatigue. The device addresses the problem of maintaining user attention during prolonged image viewing sessions by analyzing fatigue state parameters associated with individual images. These parameters may include factors such as image complexity, content novelty, or viewer interaction metrics. The device includes an interface circuit that displays an image sequence in a predefined presentation time sequence. During display, the circuit obtains a fatigue state parameter for a given image (image j) and adjusts the presentation time of subsequent images (image i) in the sequence based on this parameter. This adjustment ensures that images likely to induce fatigue are displayed for shorter durations, while others may be shown longer to sustain engagement. The system dynamically optimizes viewing time to reduce cognitive load and improve overall user experience. The invention is particularly useful in applications like digital signage, educational content delivery, or interactive displays where maintaining viewer attention is critical. The fatigue state parameter may be derived from real-time user feedback, historical data, or image analysis algorithms. The device ensures seamless integration with existing display systems while providing adaptive control over image presentation timing.
18. The image recognition device according to claim 17 , wherein the processor is configured to obtain the fatigue state parameter based on fatigue state information that is sent by a sensor and that is obtained when the observation object watches the image j.
This invention relates to an image recognition device that detects fatigue states in an observation object, such as a person, while they are viewing an image. The device addresses the problem of accurately assessing fatigue levels in real-time to improve user monitoring, safety, or performance in applications like driver monitoring, medical diagnostics, or workplace safety. The device includes a processor that analyzes an image of the observation object to determine a fatigue state parameter. This parameter is derived from fatigue state information captured by a sensor, such as an eye-tracking camera, heart rate monitor, or facial recognition system, while the object is watching a displayed image. The processor processes this sensor data to quantify fatigue, which may involve detecting eye closure rates, pupil dilation, facial expressions, or other physiological indicators. The device may also compare the fatigue state parameter against predefined thresholds to trigger alerts or adjustments, such as pausing content, adjusting display settings, or notifying a supervisor. The system enhances existing fatigue detection methods by integrating real-time sensor data with image analysis, providing a more comprehensive and dynamic assessment of fatigue. This approach is particularly useful in environments where continuous monitoring is critical, such as healthcare, transportation, or industrial settings. The device may be part of a larger monitoring system, including multiple sensors and feedback mechanisms to ensure accurate and timely fatigue detection.
19. The image recognition device according to claim 15 , wherein the processor is configured to: for the image i in the image sequence, find a presentation time corresponding to the duration impact parameter from a first mapping table, wherein the first mapping table comprises a plurality of duration impact parameters and presentation times respectively corresponding to the plurality of duration impact parameters.
This invention relates to image recognition devices that process sequences of images, particularly focusing on optimizing presentation timing based on duration impact parameters. The device addresses the challenge of dynamically adjusting how long an image is displayed to improve recognition accuracy or user experience, depending on factors like image complexity or environmental conditions. The device includes a processor that analyzes an image sequence and determines a duration impact parameter for each image. This parameter quantifies how the image's characteristics affect the required display duration. For each image in the sequence, the processor consults a first mapping table that correlates duration impact parameters with specific presentation times. The table contains multiple duration impact parameters, each linked to a corresponding presentation time, allowing the device to select the optimal display duration for each image. This ensures that images with higher complexity or other influencing factors are shown for longer, enhancing recognition performance or user engagement. The processor may also generate or update the mapping table based on historical data or real-time adjustments, ensuring adaptability to varying conditions. The system may further include additional tables or algorithms to refine the duration impact parameter calculation, such as considering environmental factors or user feedback. The overall goal is to dynamically optimize image presentation timing to improve recognition accuracy or user experience in real-time applications.
20. A non-transitory computer readable storage medium, wherein the storage medium is configured to store program instructions, which, when executed by a processor, cause the processor to perform operations of recognizing a brain-computer combination image based on image sequence presentation, the operations comprising: setting a presentation time sequence corresponding to an image sequence, wherein the image sequence comprises N images, N is a positive integer, the presentation time sequence comprises a presentation time of an image i, the presentation time of the image i is determined based on a duration impact parameter that comprises at least one of a first recognition probability calculated using a computer vision algorithm or a fatigue state parameter corresponding to the image i, and is to indicate a time period from a presentation start moment of the image i to a presentation start moment of a next adjacent image, the image i is any image in the image sequence, the presentation time sequence comprises at least two unequal presentation times, a difference between any two presentation times of the at least two unequal presentation times is k×Δ, k is a positive integer, and Δ is a preset time period value; processing the image sequence by using the computer vision algorithm, to obtain a computer vision signal corresponding to the image i in the image sequence; obtaining a feedback signal that is generated when an observation object watches the image sequence displayed in the presentation time sequence and that corresponds to the image i in the image sequence, wherein the feedback signal is to indicate a reaction of the observation object to the watched image; and fusing, for the image i in the image sequence, the computer vision signal corresponding to the image i and the feedback signal corresponding to the image i to obtain a target recognition signal of the image i in the image sequence for image recognition.
This invention relates to brain-computer interface (BCI) systems for image recognition, addressing the challenge of optimizing image presentation to improve recognition accuracy while accounting for observer fatigue and variability in recognition difficulty. The system dynamically adjusts the display duration of each image in a sequence based on a duration impact parameter, which includes either a recognition probability (calculated via computer vision) or a fatigue state parameter for each image. The presentation time sequence ensures at least two images have unequal durations, with differences between them being a multiple of a preset time period. The system processes the image sequence using computer vision to generate a computer vision signal for each image. Simultaneously, it captures a feedback signal from the observer (e.g., neural or physiological response) while they view the images. The system then fuses the computer vision signal and the observer's feedback signal for each image to produce a target recognition signal, enhancing image recognition accuracy by combining machine and human perception. This approach improves efficiency and adaptability in BCI-based image recognition tasks.
Cooperative Patent Classification codes for this invention. Click any code to explore related patents in that topic.
August 31, 2020
April 19, 2022
Browse 5M+ US patents with plain-English claim translations and AI-generated analysis.